metadata
dataset_info:
features:
- name: text
dtype: string
- name: augment_text
dtype: string
- name: best_aux
dtype: string
- name: is_improved
dtype: bool
- name: max_conditional_suffix_logp
dtype: float64
- name: original_suffix_logp
dtype: float64
splits:
- name: train
num_bytes: 4743862161
num_examples: 475258
download_size: 2366270824
dataset_size: 4743862161
configs:
- config_name: default
data_files:
- split: train
path: data/train-*